Degradation dynamics, recovery, and characterization of negative bias temperature instability M. Ershov, S. Saxena, S. Minehane, P. Clifton, M. Redford * , R. Lindley, H. Karbasi, S. Graves, S. Winters PDF Solutions, 333 West San Carlos Street, San Jose, CA 95110, USA Received 2 December 2003 Abstract This article describes several deficiencies with traditional assessments of negative bias temperature instability (NBTI) in pMOS transistors and proposes methods for handling them. These effects include: (a) a decrease in the rate of degradation over time, (b) a deviation of the stress bias dependence of NBTI lifetime from simple analytical models, (c) partial dynamic recovery of apparent NBTI degradation after interruption of stress, and (d) errors well beyond what might naively be expected in lifetime extrapolation due to uncertainties in measurement and modeling of NBTI. These errors can even be several orders of magnitude. If these effects are not adequately considered in NBTI characterization, assessment, benchmarking, and optimization, they could lead excessive expense in product reliability evaluation or, worse, to unanticipated, costly field reliability problems. Ó 2004 Published by Elsevier Ltd. 1. Introduction Negative bias temperature instability (NBTI) in pMOSFETs is a major reliability concern in modern CMOS technologies [1,2] (see also references in [2]). This degradation mechanism causes an increase of threshold voltage and decrease of the drive current, which reduces the speed of degraded transistors, increases transistor mismatch and, finally, accelerates failure of logic and analog circuits. NBTI is exacerbated by further device scaling (both vertical and lateral), increase of tempera- ture due to high dissipation power, increase of the electric field in the gate oxide, and introduction of heavy nitridation of the gate oxide. In processes with multiple gate oxides, both thin- and thick-oxide transistors suffer from NBTI. The microscopic mechanisms of NBTI are still not well understood, and optimization of process conditions to minimize NBTI is a very difficult problem. In particular, many process steps and chemical species were shown to have a strong impact on NBTI (plasma processes, back end-of-line processes, thermal budget, nitrogen, fluorine, hydrogen, etc.), but the process optimization options are often unclear or restricted. Accurate NBTI assessment is complicated by a lack of a reliable characterization procedure. Extrapolation of lifetime to nominal operating conditions based on the accelerated stress voltage measurements, and extrapo- lation of degradation data with respect to time introduce a significant uncertainty in estimated lifetime. Further complications arise due to dynamic recovery of NBTI reported recently in several papers [6–9]. NBTI degra- dation appears to be partially recoverable, which may significantly increase NBTI lifetime. On the other hand, this requires a strict control over time delays between stress interruption and monitoring measurements in characterization. In this paper, we discuss NBTI char- acterization and analysis techniques that we have found useful. A typical NBTI characterization procedure is de- scribed in ‘‘JEDEC/FSA Foundry Process Qualification * Corresponding author. Tel.: +1-408-280-7900; fax: +1-408- 280-7915. E-mail addresses: ershov@pdf.com, mark.redford@pdf.com (M. Redford). 0026-2714/$ - see front matter Ó 2004 Published by Elsevier Ltd. doi:10.1016/j.microrel.2004.03.020 Microelectronics Reliability 45 (2005) 99–105 www.elsevier.com/locate/microrel